The final project involves analyzing data about basketball players from NBA League in the 2015-2016 season.
The motivation question is :"In the 2015-2016 season, how do the skills of a player relate to his salary?"
Fall 2016
The final project involves analyzing data about basketball players from NBA League in the 2015-2016 season.
The motivation question is :"In the 2015-2016 season, how do the skills of a player relate to his salary?"
There are three kinds of raw data tables : Roster, Totals, Salaries
Steps we do here:scrape the three kinds of tables and get 90 raw data files
produce roster-salary-stats.csv file, which contains all variables from Roster,
Totals and Salary.
Picture of code we use to scrape and download data
Two aspects of analysis : Qualitative amd Quantitative
Qualitative variables:player's name, birth date, country, team, position, etc.
Quantitative variables: mean, median, min, standard deviation, etc.
bar chart of counts of positions
bar chart of counts of number of players in each team
We can see that the highest frequency of salary centered around 2000000.
This bar chart shows the total payroll for each team in ascending order.
This bar chart shows the first quantile of salary statistics for each team in descending order
Old Eff formula = (PTS + REB + AST + STL + BLK - Missed FG - Missed FT - TO) / GP
Only favors offense-oriented players. Not good!
New EFF = w1x1/std1 + …w8x8/std8, where xj is the j-th variable, j = 1,2,…,8, wj is the weight for variable j = 1,2,…8, stdj is the standard deviation of variable j = 1,2,…,8
Positive relation between total points and salary
The picture shows the relation between salary and missed free throws.
Formula: value = efficiency/salary. We want to identify the top 20 players (most valuable) and the bottom 20 players(worst value)
From our analysis, although there is a positive relation between players' salary and their efficiency, there are some players who have high salaries but missed a lot of field goals, and some players who earn average or low salaries actually perform well. So basketball teams should use an accurate statistical method such as PCA and modified EFF formula to analyze players' performances as one of standards to determine their salary.